This application responds to PAR-13-080 - Accelerating the Pace of Drug Abuse Research Using Existing Data. Functional magnetic resonance imaging (fMRI) and the traditional general-linear-model-based analysis (GLM-BA) have been used to assess neural correlates of cocaine dependence (CD) and other neuropsychiatric disorders. However, these studies have reported inconsistent findings including reduced and increased brain activity in patients relative to healthy controls (HCs), and have not been able to provide a clear picture on CD neuropathology and define reliable biomarkers for CD treatment. Investigators have suggested multiple factors such as addiction stages of patients contributing to conflicting findings. We recently applied spatial independent component analysis (sICA) to several fMRI datasets and found extensive overlaps of functional networks (FNs) with opposite timecourses of task-related activities. Based on these findings, we predict: 1) that blood-oxygenation-level-dependent (BOLD) signal mixtures reflecting concurrent co-localized activation and deactivation (CCAD) of different neurons in the same voxels contribute to the opposite findings, while CCAD is determined by balanced excitation and inhibition and functional heterogeneity in the brain; 2) that due to the nature of signal mixtures,a reduced brain deactivation might be expressed as an increased BOLD signal and misinterpreted as a greater brain activation by studies using GLM-BA, which cannot differentiate mixed signals from same voxels; and 3) that sICA will overcome the limitation of GLM-BA and reveal consistent fMRI findings in patients, because sICA can separate signal mixtures from the same voxels into different FNs, and thus mitigate the confusion of reduced deactivation vs. increased activation. For testing our predictions and better understanding CD neuropathology, this project will use sICA to perform secondary analyses on fMRI data acquired from 419 participants, including 179 HCs and 196 CD patients. They performed fMRI tasks for assessing cognitive control and/or monetary reward/loss- motivation. Most patients were treated for CD and followed for 3 ~ 12 months after treatment. We hypothesize that sICA will consistently reveal reduced task-related modulations of FNs related to cognitive control and/or reward/loss-motivation in CD patients relative to HCs, increased (i.e., recovery) modulations of these FNs in patients after effective treatment for CD, and positive correlations between greater modulations of these FNs and better long-term outcomes after treatment. Findings supporting our hypotheses will not only provide a consistent insight into CD neuropathology and reconcile extant conflicting data, but also demonstrate that task- related modulations in FNs related to cognitive control and/or reward/loss-motivation in CD patients as revealed by sICA are excellent candidates for reliable diagnostic and predictive biomarkers for optimizing and developing CD treatments. Furthermore, these findings will have a sustained, powerful impact on fMRI theories and practices in a general and broad sense and will help move the entire field forward.